You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Many of our fix messages are redundant, obvious, or otherwise unhelpful. We should use the fix feature more mindfully. That means supplying the "corrected" code or adding something to the diagnostic that isn't already conveyed. And if those can't be accomplished, the fix should be omitted.
note[BlankLinesBetweenElements]: missing blank line
┌─ tests/lints/blank-lines-between-elements/source.wdl:17:5
│
17 │ ╭ scatter (i in ["hello", "world"]) {
18 │ │ call bar { input: s = i }
19 │ │ }
│ ╰─────^
│
= fix: add a blank line before this element
We discussed being more intentional with the fix messages. So we probably want to drop this one as it doesn't add very much. That leaves:
note[BlankLinesBetweenElements]: missing blank line
┌─ tests/lints/blank-lines-between-elements/source.wdl:17:5
│
17 │ ╭ scatter (i in ["hello", "world"]) {
18 │ │ call bar { input: s = i }
19 │ │ }
│ ╰─────^
Problem here is that now the Span doesn't make sense. Why are lines 18-19 included? Maybe the span's len should be 0? That creates:
note[BlankLinesBetweenElements]: missing blank line
┌─ tests/lints/blank-lines-between-elements/source.wdl:17:5
│
17 │ scatter (i in ["hello", "world"]) {
│ ^
Ideally, I want an arrow that points between lines 16 and 17 and indicates a blank goes here. IDK how to do that in codespan. Can we do that?
Created by hand, but something like:
┌─ tests/lints/blank-lines-between-elements/source.wdl:17:5
16 │ input {}
---->
17 │ scatter (i in ["hello", "world"]) {
(We can't do that in codespan)
The moral of this experiment is that reworking the fix messages may also require changing up some of the reported Spans to make more sense. That can be a bit tricky depending on how the rule is written. So just something to be mindful of.
Many of our
fix
messages are redundant, obvious, or otherwise unhelpful. We should use thefix
feature more mindfully. That means supplying the "corrected" code or adding something to the diagnostic that isn't already conveyed. And if those can't be accomplished, thefix
should be omitted.related to #192
The text was updated successfully, but these errors were encountered: